Fidelity-preserving coefficient truncation method for wavelet-based compressors for biomedical IoT applications

Jose Santos, Dongming Peng, Michael Hempel, Hamid Sharif

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

A common approach in the realization of wavelet-based compressors for ECG signals makes use of truncation methods, whereby wavelet coefficients are truncated (i.e., thrown away) if they're deemed insignificant. A popular truncation strategy applied in these compressors is to truncate based on Energy Packing Efficiency or EPE, which tends to favor coefficients at higher scales because their energy contribution (either in squared or absolute-value sense) is itself insignificant. In this paper, we present four rudimentary truncation strategies to analyze and demonstrate how the choice of truncation strategy can affect the signal in terms of compression ratio (CR) and signal fidelity. Of these, a truncation strategy we call 'ScaleRelativeMAX' is proposed, which exhibits some useful properties for sensitive biomedical applications. Simulation results are presented using representative select ECG records from PhysioNet's database to show that some truncation methods-in particular, our proposed truncation strategy-allows for fine-grained fidelity and CR control than others and offer nearly linear reconstruction error growth as a function of the truncation threshold in comparison to other strategies that are more aggressive in favoring CR over signal fidelity. Such fidelity-first strategies are useful in biomedical communication architectures for emerging Internet-of-Things (IoT) applications that employ compressors to minimize energy transmission costs in Body Area Sensor Networks (BASNs) and similar wearable devices. Such signals carry diagnostic information that are of clinical significance, and whose reconstruction of clinical features should take priority and is in stark contrast to ordinary multimedia class signals, which generally tend to favor CR over signal fidelity.

Original languageEnglish (US)
Title of host publication2017, 11th International Conference on Signal Processing and Communication Systems, ICSPCS 2017 - Proceedings
EditorsBeata J Wysocki, Tadeusz A Wysocki
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1-8
Number of pages8
ISBN (Electronic)9781538628874
DOIs
StatePublished - Jan 25 2018
Event11th International Conference on Signal Processing and Communication Systems, ICSPCS 2017 - Surfers Paradise, Australia
Duration: Dec 13 2017Dec 15 2017

Publication series

Name2017, 11th International Conference on Signal Processing and Communication Systems, ICSPCS 2017 - Proceedings
Volume2018-January

Other

Other11th International Conference on Signal Processing and Communication Systems, ICSPCS 2017
CountryAustralia
CitySurfers Paradise
Period12/13/1712/15/17

Keywords

  • Architectures
  • BASN
  • Biomedical
  • Body Area Sensor Networks
  • Communication
  • Compressors
  • ECG
  • Energy-Packing Efficiency
  • Fidelity-Preserving
  • IoT
  • PhysioNet
  • Strategies
  • Truncation
  • Wavelet
  • Wearables

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Signal Processing
  • Instrumentation

Fingerprint Dive into the research topics of 'Fidelity-preserving coefficient truncation method for wavelet-based compressors for biomedical IoT applications'. Together they form a unique fingerprint.

  • Cite this

    Santos, J., Peng, D., Hempel, M., & Sharif, H. (2018). Fidelity-preserving coefficient truncation method for wavelet-based compressors for biomedical IoT applications. In B. J. Wysocki, & T. A. Wysocki (Eds.), 2017, 11th International Conference on Signal Processing and Communication Systems, ICSPCS 2017 - Proceedings (pp. 1-8). (2017, 11th International Conference on Signal Processing and Communication Systems, ICSPCS 2017 - Proceedings; Vol. 2018-January). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICSPCS.2017.8270509